Murmur vs VoiceNPC: Which Local Voice App Fits Your Work?

The short answer

Choose VoiceNPC if you want fast on-device cloning across iPhone, iPad, and Mac, built-in translation, cross-language output, a free ad-supported evaluation, or one script generated in several languages. Choose Murmur if the Mac is your production desk and you need several local voice model families, 860+ community voices, Voice Design, a batch queue, multi-script projects, reusable speakers, timeline editing, and organized audio export. VoiceNPC is the more flexible Apple-device cloner. Murmur is the broader Mac production studio. Both keep voice and script processing local, but neither should be described as making no network requests at all. Facts and US price examples in this comparison were checked against official sources on July 12, 2026.

In this comparison

• Current pricing and free access
• Mac, iPhone, and iPad requirements
• Voice sample length and model choice
• Translation, languages, and multi-voice projects
• Privacy and what local really means
• Export and long-form production workflow
• Who should choose each app

Murmur vs VoiceNPC at a glance, checked July 12, 2026

VoiceNPC prices vary by storefront. The current US App Store listing uses the prices above. The VoiceNPC website presents them as early-bird discounts from higher regular prices, while other countries show different amounts. Check the storefront tied to your Apple ID before comparing lifetime totals.

The real difference is what happens after the clone

A short clone is easy to demonstrate. Production is harder. A creator has to organize scripts, assign speakers, regenerate one bad line, compare takes, preserve timing, and export files that work in an editor. VoiceNPC and Murmur approach that work from different directions. VoiceNPC begins with a portable voice identity and makes it easy to reuse that voice across Apple devices and languages. Murmur begins with a Mac project and brings voices, scripts, clips, and exports into one desktop workspace.

That is why this is not a simple quality ranking. Someone localizing an app demo while traveling may value VoiceNPC's translation flow more than a desktop timeline. Someone producing a ten-chapter course may value Murmur's queue, project structure, and model choice more than mobile access. Test the job, not just the first sentence.

Voice cloning and voice choice

VoiceNPC documents a clear 5 to 15 second input range. You can record directly, import audio, or extract a voice sample from video. The clone becomes a reusable local voice fingerprint. VoiceNPC also ships with eight starter voices, so a first generation does not require a recording. Its current product is built around Qwen3-TTS running locally with style presets such as Stable, Natural, and Expressive.

Murmur's public site markets cloning from a 10-second clip. The current dedicated clone importer accepts 10 to 210 seconds and marks 30 seconds as a good target, while another save-to-clone flow recommends roughly 5 to 10 seconds. The practical advice is to use one clean speaker with little room noise and enough natural variation to represent the voice. The local voice-cloning tutorial covers recording preparation.

Murmur offers more ways to avoid cloning entirely. Its catalog includes Kokoro, two Chatterbox variants, Qwen3-TTS, SparkTTS, Fish Audio S2 Pro, OmniVoice, and a Qwen Voice Design workflow. The Murmur voice library lists 860+ community choices. That breadth helps when a cloned voice works for a host but not a character, narrator, or second speaker. The tradeoff is model downloads, more choices to understand, and higher storage or memory needs for larger engines.

Translation and multilingual work

VoiceNPC has the clearer localization workflow. Type a script in one supported language, choose a target language, preview the on-device translation, edit it, and synthesize the translated text in the cloned voice. Free users handle one target language at a time. Pro users can select several targets and produce a separate generation for each in one pass. Translation uses Apple's local Translation framework after the required language packs are downloaded.

VoiceNPC lists English, Chinese, Japanese, Korean, French, German, Spanish, Russian, Italian, and Portuguese, plus Beijing and Sichuan dialect support. It also supports mixed-language scripts and cross-language cloning. This is a meaningful advantage if your source voice records in English but your audience needs Spanish, Japanese, or Portuguese.

Murmur supports multilingual speech, but language availability depends on the selected engine. Chatterbox Multilingual, Qwen3-TTS, Fish Audio, Kokoro, and OmniVoice each expose different language sets. Murmur does not translate the script for you. You must provide the text in the output language and verify it yourself. Do not confuse multilingual synthesis with translation. The Murmur languages page helps match a language to a model.

Projects, speakers, and revision workflow

VoiceNPC supports multi-voice projects with a rich section editor and a markup mode. Each section can use a different voice and language. You can edit and regenerate individual sections, play the result as a playlist, and generate the whole project. That is a strong fit for short dialogues, localized videos, podcasts, and audiobooks assembled section by section.

Murmur's Projects workspace goes further into desktop production. A project can hold several scripts and a shared speaker roster. Speakers can use preset, cloned, community, or designed voices. Generated segments become clips on a timeline with takes, trims, fades, gain, mute state, markers, and media lanes. You can render a full timeline or export clips separately. A preflight check can flag unassigned voices, stale clips, and muted material before export.

VoiceNPC is simpler and more portable. Murmur is deeper and more demanding. If your final edit already happens in Final Cut Pro, DaVinci Resolve, Logic, or another desktop editor, Murmur's local project and stem workflow can remove repetitive file handling. If you mainly need several voiced sections and a share sheet, VoiceNPC may be enough.

Privacy: both are local, but the details differ

VoiceNPC says voice cloning, translation, and speech synthesis happen on-device, with voice fingerprints stored in the app sandbox. That protects the content of the reference voice and script from hosted inference. However, the App Store privacy label says identifiers and usage data may be used to track users across apps and websites, and the free tier contains advertising. The accurate claim is that voice and script processing remain local, not that the whole app is network-free.

Murmur also performs generation locally. Its privacy policy says it does not collect typed text, generated audio, full prompts, documents, files, microphone data, or camera data. It does disclose license validation, aggregate feature telemetry, and crash diagnostics. The Murmur home page still contains stronger no-telemetry wording, but the dated privacy policy is the safer source. Neither product's local architecture removes the need to obtain consent before cloning another person.

Pricing and free access

Murmur costs $49 once, has no free trial, and includes a seven-day refund policy. There are no generation credits or monthly character caps. The cost is predictable, but a buyer has to pay before evaluating the app on a real script. Murmur makes the most financial sense when voice generation is recurring work and the production features replace several smaller tools.

VoiceNPC is easier to test. Its free tier includes 25 energy each day, eight starter voices, all listed languages, unlimited saved clones, multi-voice projects, and one translation target at a time. One energy is described as roughly 100 characters. Users can earn more through ads, daily sharing, and referrals. Pro removes ads and generation limits and adds one-pass multi-language output. Current US pricing is $2.99 monthly, $29.99 annually, or $59.99 lifetime, but the developer labels this an early-bird offer.

If the question is only which product is easier to evaluate, VoiceNPC wins. If the question is whether a $49 purchase can become the center of a weekly Mac narration workflow, Murmur has the broader case.

Export and system requirements

VoiceNPC exports WAV and lets users share a voice fingerprint as a .voicenpc file through AirDrop, Messages, or another share target. It requires iOS 26, iPadOS 26, or macOS 26. Supported hardware starts at iPhone 13 and selected newer iPads, while the Mac version requires Apple Silicon. The App Store download is roughly 2 GB before any additional language resources.

Murmur is Mac-only and Apple Silicon-only. The current build target is macOS 15, even though older public pages still say macOS 14. Current app code exposes WAV and M4A output, including full timeline renders and separate clips. Some public copy says MP3, but buyers should not rely on MP3 until the shipped app exposes it. Murmur also needs one-time model downloads, and larger models benefit from more than the 8 GB minimum advertised on the site.

Who should choose Murmur?

• Mac creators producing narration, courses, podcasts, dialogue, or audiobooks every week
• People who want several local voice model families rather than one main engine
• Projects that need reusable speakers, multiple scripts, queueing, takes, timeline edits, or clip stems
• Buyers who prefer one $49 purchase to subscriptions or ad-based energy
• Creators who want a large preset and community voice catalog alongside cloning

Who should choose VoiceNPC?

• Creators who need the same cloning workflow on iPhone, iPad, and Mac
• People translating and voicing one script in several supported languages
• Buyers who want to evaluate cloning for free before purchasing
• Projects that benefit from shareable voice fingerprints
• Users who prefer a simpler section-based project over a desktop production timeline

A fair five-minute test

1. Record one clean 15-second sample with a natural pace and no background music.
2. Generate the same 120-word paragraph in both apps.
3. Change one difficult name and regenerate only the affected section.
4. Add a second speaker and listen to the transition between voices.
5. Export the audio and place it in the editor you actually use.
6. If localization matters, translate one paragraph in VoiceNPC and provide your own translated version to Murmur. Have a fluent speaker review both texts before judging the voices.

Frequently asked questions